Version 1.0, March 1st, 2018
By: Dr. Elad Harison, Chief Architect and COO, IAGON
Introduction
The rapid growth of introducing new crypto-token projects and ICOs, virtually in every domain of business, industrial, IT and public operations creates wealth of innovation and inventive solutions that are based on the blockchain technology. Nonetheless, the forms in which information of each project, its roadmap and scope is disclosed require vast efforts from investors and community members attempting to identify the properties of each project, its direction of development, roadmap, operations and support.
IAGON operates in the field of decentralized cloud services, developing a platform (i.e. a “grid”) for conducting processing tasks and storing files to cater to the needs of both conventional computational tasks and databases, as well as to the needs of the more complex AI operations and management of Big Data repositories that are significantly larger than organizational databases.
IAGON conducted this study to assess the capabilities of other crypto-token and blockchain based projects that address decentralized storage of processing, and completed their ICO or plan to do so. These projects were compared according to prominent features of cloud services, decentralized storage, processing and the existing crypto-token platforms. Data was collected and analyzed from multiple resources published by different projects: whitepapers, websites, videos, forums, project blogs, etc. Despite our efforts, no references to particular attributes of some projects were found online. We will be grateful for completing these details, including references from the originators of the projects, for the benefit of the blockchain, Big Data and AI community members.
The Criteria for Benchmarking
Decentralized cloud services provide both storage and processing capacity and as the recent emergence of AI technologies require not only CPU use but also GPU, the following dimensions were included in our analysis:
AI powered grid — Does the processing/storage grid utilize advanced algorithms, such as Machine Learning techniques (data classification and clustering), Artificial Neural Networks or Deep Learning to continuously analyze its performance, improve and optimize it? The implementation of an AI powered grid can assist in the allocation of processing tasks and files for storage based on the current and predicted miners’ performance and accessibility.
Decentralized management of the grid — Is the processing/storage planned to be managed by a decentralized system for the allocation of processing tasks and files for storage by using multiple nodes, or is the management of the grid based on a central module that allocates them to the miners?
Continuous optimization of the grid — Is the grid planned to “learn” from its present performance by continuously analyzing data on it and dynamically modifying the allocation of processing tasks and files for storage to optimize it under changing conditions, or does it use a static and non-adaptive policy (such as random allocation)?
Storage of file fragments on the grid
CPU processing grid — Is the grid planned to support conventional processing tasks that are based on the use of CPUs (such as SQL queries)?
GPU processing grid — Is the grid planned to support conventional processing tasks that are based on the use of GPUs (such as Deep Learning analysis)?
Multiple platform support — Is the grid planned to support other crypto-platforms, in addition to Ethereum?
Ethereum platform — Is the grid planned to run on Ethereum?
Tangle platform — Is the grid planned to run on the Tangle technology?
OS support — Which operating systems will the miner’s application support?
File encryption — Which file encryption algorithm is planned to secure stored files?
Results
The benchmarking analysis was based on a comparative study of projects developing decentralized storage and processing platforms that completed or plan to complete an ICO. Data was collected from their whitepapers, websites, videos, forums, project blogs, as well as from other materials (such as FAQ and replies to questions) published online by the project teams.
The results of the comparison considering the multiple dimensions of comparison are as follows:
IAGON
IAGON’s architecture is planned to support allocation of file fragments for storage and processing tasks operated on a decentralized grid of nodes that join the storage and processing capacities of miners. The allocation of processing tasks and files for storage will be done by operation of an AI based coordination module that will operate in a decentralized manner. It will utilize the miner’s applications on the various nodes to select the miner’s node for carrying out a processing task or for storing a file fragment. The project will support multiple technologies, including Ethereum and Tangle.
Storj
Storj is planned to support the storage of files on a decentralized grid by using Ethereum. The management of the platform is decentralized and miner’s applications will run on different operating systems. Files are protected by using the AES256 protocol to secure them before allocating them to miners.
SONM
SONM is planned to support both storage of files and processing tasks on a decentralized grid by using Ethereum. The management of the platform is decentralized and miner’s applications will run on different operating systems. The encryption of files stored on miners’ nodes is not clearly indicated, though SONM’s whitepaper mentioned that “Hard disk drive analogue will be implemented using decentralized data storage solutions: IPFS (InterPlanetary File System), Storj, Sia, etc.”. SONM will support processing tasks on both CPUs and GPUs.
DADI
DADI is planned to support both storage of files and processing tasks on a decentralized grid by using Ethereum. The management of the platform is decentralized and miner’s applications will run on different operating systems. The encryption of files stored on miners’ nodes is not clearly indicated. DADI clearly indicates that it will support processing tasks on CPUs, but does not mention whether processing on GPUs will be included in its scope of operations.
Golem
Golem is planned to support processing tasks on a decentralized grid by using Ethereum. The management of the platform is decentralized and miner’s applications will run on different operating systems. Golem will support processing tasks on CPUs. Support of processing tasks that require GPU operations will be announced by Golem’s team.
iExec
iExec is planned to support both storage of files and processing tasks on a decentralized grid by using Ethereum. Management of the platform is decentralized and miner’s applications will run on different operating systems. The encryption of files stored on miners’ nodes is not clearly indicated. iExec will support processing tasks on both CPUs and GPUs.
dfinity
dfinity is planned to support both storage of files and processing tasks on a decentralized grid by using Ethereum. Management of the platform is decentralized and miner’s applications will run on different operating systems, though dfinity’s team does not indicate which operating systems will be supported. The encryption of files stored on miners’ nodes is not clearly indicated. dfinity will support processing tasks on both CPUs and GPUs.
Siacoin
Siacoin is planned to support storage of files on a decentralized grid by using Ethereum. Management of the platform is decentralized and miner’s applications will run on different operating systems. Files are protected by using the blake2b protocol to secure them before allocating them to miners.